tflite2tensorflow v1.22.0

• Add tensor unit quantization option to tflite quantization options (experimental implementation)
  • --disable_per_channel

tflite2tensorflow v1.21.2

• Security update
• README update

tflite2tensorflow v1.21.1

• CUDA 11.6
• TensorRT 8.4.0
• cuDNN 8.4
• TensorFlow v2.9.0
• TensorFlow Lite v2.9.0
• PyTorch v1.12.0 (with grid_sample)
• onnxruntime-gpu v1.12.0

tflite2tensorflow v1.21.0

• CUDA 11.6
• TensorRT 8.4.0
• cuDNN 8.4
• TensorFlow v2.9.0-rc0
• TensorFlow Lite v2.9.0-rc0
• PyTorch v1.11.0

tflite2tensorflow v1.20.9

https://github.com/PINTO0309/simple-onnx-processing-tools

1. sne4onnx
2. snd4onnx
3. snc4onnx
4. scs4onnx
5. sog4onnx
6. sam4onnx
7. soc4onnx
8. scc4onnx

tflite2tensorflow v1.20.8

• Support for conversion of face_landmark_with_attention to OAK
  • issue: Add option to optimize tf.math.reduce_prod to Myriad (OAK) #29

tflite2tensorflow v1.20.7

• Added Unity Barracuda support option for GatherND (gather_nd)
  • --optimizing_barracuda
  • Generates ONNX by replacing Barracuda unsupported layers with standard layers. For example, GatherND

tflite2tensorflow v1.20.6

• Support for MediaPipe Custom OP transform_landmarks,transform_tensor_bilinear,landmarks_to_transform_matrix

tflite2tensorflow v1.20.5

• Fixed NCHW conversion bug on ONNX export
• Remove NNPACK
• Update schema.fbs

tflite2tensorflow v1.20.4

• Support for transform_landmarks, transform_tensor_bilinear, landmarks_to_transform_matrix
  • tensorflow-2.8.0-cp38-none-linux_x86_64.whl
  • tflite_runtime-2.8.0-cp38-none-linux_x86_64.whl